Hi, An old system I had which I wanted to resurrect. I remember back in the days (2003-2004), linux support for the vt1211 was a bit tough (both in terms of driver & formulas for lm-sensors). Come to think of it, it still is cumbersome, as neither "SpeedFan" or "Everest" (and others I tried) in windows can reflect sensible sensor information. M/B: Foxconn 600A01-6LRS CPU: Athlon XP 2800+ Barton (FSB 333Mhz, Clock 2.075Ghz) # lshw [..snip..] description: Motherboard product: KT600-8237 # lspci -n [..snip..] 00:11.0 0601: 1106:3227 according to http://www.kernel.org/doc/Documentation/i2c/busses/i2c-viapro 1106:3227 is VT8237R # dmesg | grep vt1211 [ 18.413205] vt1211: Found VT1211 chip at 0x6000, revision 2 # sensors-detect # sensors-detect revision 5818 (2010-01-18 17:22:07 +0100) # System: VIA Technologies, Inc. KT600-8237 [..snip..] Found `VIA VT1211 Super IO Sensors' Success! (address 0x6000, driver `vt1211') [..snip..] Using driver `i2c-viapro' for device 0000:00:11.0: VIA Technologies VT8237 South Bridge $ sensors acpitz-virtual-0 Adapter: Virtual device temp1: -273.2°C (crit = +60.0°C) vt1211-isa-6000 Adapter: ISA adapter in2: +1.69 V (min = +1.60 V, max = +1.70 V) in3: +1.94 V (min = +1.92 V, max = +2.25 V) in4: +2.22 V (min = +1.89 V, max = +2.32 V) +3.3V: +3.17 V (min = +3.13 V, max = +3.46 V) fan1: 3971 RPM (min = 3504 RPM, div = 2) fan2: 3600 RPM (min = 3006 RPM, div = 2) temp1: +40.0°C (high = +255.0°C, hyst = +0.0°C) SIO Temp: +51.0°C (high = +65.0°C, hyst = +60.0°C) temp3: +0.7°C (high = +0.5°C, hyst = +0.5°C) temp4: +0.9°C (high = +0.5°C, hyst = +0.6°C) cpu0_vid: +1.850 V ------ in2 is Core Voltage (no calc needed, so looks good) in3, in4 I suppose are +5V and +12V, though way off fan1 is CPU Fan fan2 is System Fan temp3 seems to be CPU Temp temp4 would be M/B Temp base temperature values, idle (@room temp 25.2C, case temp 28.2C): $ grep "" /sys/devices/platform/vt1211.24576/temp*_input /sys/devices/platform/vt1211.24576/temp1_input:40000 /sys/devices/platform/vt1211.24576/temp2_input:51000 /sys/devices/platform/vt1211.24576/temp3_input:733 /sys/devices/platform/vt1211.24576/temp4_input:859 # burnK7 (from CPUBurn - ran for a few minutes until values stabilize) $ grep "" /sys/devices/platform/vt1211.24576/temp*_input /sys/devices/platform/vt1211.24576/temp1_input:40000 /sys/devices/platform/vt1211.24576/temp2_input:51000 /sys/devices/platform/vt1211.24576/temp3_input:576 /sys/devices/platform/vt1211.24576/temp4_input:828 After a bit of research and testing, I made my config file: /etc/sensors.d/foxconn_600A01-6LRS and ran `sensors -s` now the equivalent output of `sensors` for the two runs above is: IDLE (room temp 25.2C, case temp 28.2C): $ sensors acpitz-virtual-0 Adapter: Virtual device vt1211-isa-6000 Adapter: ISA adapter VCore: +1.69 V (min = +1.60 V, max = +1.70 V) +5V: +4.76 V (min = +4.61 V, max = +5.41 V) +12V: +12.67 V (min = +10.77 V, max = +13.21 V) +3.3V: +3.17 V (min = +3.13 V, max = +3.46 V) CPU Fan: 3996 RPM (min = 3504 RPM, div = 2) SYS Fan: 3581 RPM (min = 3006 RPM, div = 2) SIO Temp: +50.0°C (high = +65.0°C, hyst = +60.0°C) CPU Temp: +43.5°C (high = +64.8°C, hyst = +60.2°C) SYS Temp: +36.4°C (high = +60.2°C, hyst = +55.2°C) cpu0_vid: +1.850 V Full load w/ burnK7 for several minutes till peak temp is reached: $ sensors acpitz-virtual-0 Adapter: Virtual device vt1211-isa-6000 Adapter: ISA adapter VCore: +1.69 V (min = +1.60 V, max = +1.70 V) +5V: +4.71 V (min = +4.61 V, max = +5.41 V) +12V: +12.61 V (min = +10.77 V, max = +13.21 V) +3.3V: +3.15 V (min = +3.13 V, max = +3.46 V) CPU Fan: 4096 RPM (min = 3504 RPM, div = 2) SYS Fan: 3561 RPM (min = 3006 RPM, div = 2) SIO Temp: +52.0°C (high = +65.0°C, hyst = +60.0°C) CPU Temp: +54.5°C (high = +64.8°C, hyst = +60.2°C) SYS Temp: +38.7°C (high = +60.2°C, hyst = +55.2°C) cpu0_vid: +1.850 V And here's my config, heavily commented as to give best picture possible, and hopefully to be corrected wherever I'm wrong :) ####### Begin /etc/sensors.d/foxconn_600A01-6LRS ###### chip "acpitz-virtual-0" ignore temp1 # returns bogus value -273200 (of course, no sensor present) chip "vt1211-*" ignore temp1 # seems to be grounded (always 40000) label in2 "VCore" label in3 "+5V" label in4 "+12V" label in5 "+3.3V" label temp2 "SIO Temp" # (SuperI/O Chip; VT1211) label temp3 "CPU Temp" label temp4 "SYS Temp" label in2 "VCore" label in3 "+5V" label in4 "+12V" label in5 "+3.3V" label fan1 "CPU Fan" label fan2 "SYS Fan" # All voltage calculations have the form # compute inX @ * (1 + R1 / R2), @ / (1 + R1 / R2) # # The following are the resistor values as recommended by VIA # Voltage R1 R2 # ------- ---- ---- # VCore - - (no scaling) # 3.3 6.8k 10k (R42 & R50, respectively) # 5.0 14k 10k (R504/R0603 & R50, respectively) # 12.0 47k 10k (R43 & R51, respectively) # # The VT1211 internal 3.3V (in5) is scaled by the driver and doesn't # need to be adjusted here. VCore (in2) doesn't need scaling at all. compute in3 @ * (1 + 14 / 10), @ / (1 + 14 / 10) compute in4 @ * (1 + 47 / 10), @ / (1 + 47 / 10) # `cat /sys/devices/platform/vt1211.24576/cpu0_vid` gives 1850 (~ 1.850V) which is wrong # Athlon 2800+ Barton Core (333Mhz FSB) VCore is 1.65V. Adjust for your CPU, if needed. set in2_min 1.65 * 0.97 set in2_max 1.65 * 1.03 # For the 5V input (in3), at least, you may want to widen the acceptable rage a bit # (by increasing the multiplier of in3_max, and decreasing the multiplier of in3_min) # this is because the manufacturer's own schematic for this motherboard clearly shows # +/-1% for R504 (14K_NC); the resistor on 5V_SYS that connects (along with 3D3V_SYS) # to UIC4 (in lm-sensors/kernel-2.6 terms - UCH4) # For the schematic, google "600a01-c-hw-sch-final.pdf" (page 30 of 37) # If you happen to have windows installed, download the "SuperUtility" software for # this board from foxconn site, run the bundled "SuperStep" utility and look at how # wide of min/max ranges they have set as default. For example: the 5V input has its # alarm limits set as 4V for min, 6V for max! (a little ridiculous) set in3_min 5.0 * 0.92 set in3_max 5.0 * 1.08 set in4_min 12.0 * 0.90 set in4_max 12.0 * 1.10 set in5_min 3.3 * 0.95 set in5_max 3.3 * 1.05 # The VT1211 internal temperature (temp2) is scaled by the driver # and doesn't need to be adjusted here. # The thermistor calculations are of the form # compute tempX 1 / (1 / 298.15 - (` Vmax / @ - 1)) / B) - 273.15, \ # Vmax / (1 + (^ (B / 298.15 - B / (273.15 + @)))) # # B is the thermistor beta value, Vmax is the reference voltage, '^' is the # exp() operator and '`' is the ln() operator. # Given B = 3435 and Vmax = 2.2V and assuming that the thermistor forms a # resistor divider with a resistor equal to the thermistor's nominal value at # 25 degrees C, the following compute lines can be used: # For more info, see: http://www.kernel.org/doc/Documentation/hwmon/vt1211 # Please note that results *will be noticably different* from what you might see in BIOS, # and even what you see in "SuperStep" (from "SuperUtility") if you happen to run windows, # even when comparing idle temps, but that doesn't mean what you get here is wrong! Why? # 1. Page 30 of the schematic (mentioned above) clearly states: # "MONITOR CPU TEMP BY RT1. UIC1:CPU TEMP, RT1 Placed in CPU cavity." # That certainly means a thermistor measuring "surface temp", NOT "core temp". # 2. BIOS alarm setting for CPU temp offers 4 options; 60C, 70C, 80C, and disabled. # Considering that Athlon XP CPUs "Internal Fry Temp" is 85C, which corresponds to a # "Surface Fry Temp" of 75C, the BIOS cannot possibly be reflecting surface temps when # alarm max setting is @ 80C (actual shutdown occurs at or around 82-84C). # Therefore, the BIOS (and "SuperStep" utility) adds to the thermistor-reported temps. # To confirm, conduct a simple "Cold Power-up Test": # 1. Let machine cool down to room temperatue (preferably overnight), switch on, go straight # to the BIOS -> "PC Health" to check the temps. At the instant you power-on, system and # CPU temps should both equal Room temp. Any significant difference is a calibration # error (or intentional difference/offset) in the readouts. # 2. Note how long it takes for temp to rise and stabilize at max Idle Temp. It should take # anywhere from 10-15 minutes to level off, indicating the test itself is OK. # In my test, *cold boot* (overnight) and straight to BIOS, CPU/SYS temps were 32C/32C, # while the room and case temps both measured exactly at 24.2C (two different thermostats). # Calibration difference = 32 - 24 = 8C , give or take 1C. Therefore, offset is 7C to 9C. # If you compare your CPU or SYS idle temp from `sensors` output (per the calculation below) # with its counterpart in the BIOS, you'll find same difference range of 7C to 9C. # For example, with room temp @ 24.2C, and case temp @ 28.2C, I get: # `sensors` reports: CPU 43.5C & SYS 36.4C, occasional slight shift to 44.1C & 37.0C # BIOS temp reports: CPU 51C & SYS 43C (both BIOS & SuperStep reported temps are integers) # SuperStep reports: CPU 51C & SYS 43C, occasional slight shift to 51C-52C & 44C compute temp3 1 / (1 / 298.15 - (` (2.2 / @ - 1)) / 3435) - 273.15, \ 2.2 / (1 + (^ (3435 / 298.15 - 3435 / (273.15 + @)))) compute temp4 1 / (1 / 298.15 - (` (2.2 / @ - 1)) / 3435) - 273.15, \ 2.2 / (1 + (^ (3435 / 298.15 - 3435 / (273.15 + @)))) set temp2_max 65 # SIO (internal temp) ALRM set temp2_max_hyst 60 # SIO (internal temp) ALRM clear set temp3_max 65 # CPU (surface temp) ALRM set temp3_max_hyst 60 # CPU (surface temp) ALRM clear set temp4_max 60 # SYS (surface temp) ALRM set temp4_max_hyst 55 # SYS (surface temp) ARLM clear set fan1_min 3500 # CPU FAN set fan2_min 3000 # SYS FAN ####### End /etc/sensors.d/foxconn_600A01-6LRS ###### (I have the motherboard schematic. Was a bit hard to find - had to register to chinese search engine; baidu, to be able to download it. Can upload it somewhere else upon request) Please let me know if you think it looks okay, and whether it's worth posting in the "motherboard specific configuration" section of the wiki. Thanks, Jasmine _______________________________________________ lm-sensors mailing list lm-sensors@xxxxxxxxxxxxxx http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
